By Per-Anders Edin, Tiernan Evans, Georg Graetz, Sofia Hernnäs and Guy Michaels
Technological change facilitates long run growth, but labour-replacing technologies are often perceived as a threat to the livelihood of workers. The recent wave of technologies is no longer confined to replacing routine workers such as machine operative and clerical workers: it could well reduce the employment of, among others, drivers, lawyers, and even fashion designers. Since more occupations are likely to be affected, it is important to understand how costly it is for workers, since this informs our thinking about individual welfare, inequality, and human capital investments.
In a new study, we examine the lifetime earnings and employment consequences for workers when demand for their occupation falls. We use administrative data on all Swedish workers to identify those who in 1985 worked in occupations that subsequently declined. We follow these workers’ careers, including earnings, employment, unemployment, and retraining spells, over almost 30 years. We compare workers affected by occupational decline to workers who were initially very similar in terms of their demographic characteristics, schooling, and earnings. We show that after adjusting for these attributes, workers who faced occupational decline were no different from others in terms of their initial cognitive and non-cognitive skills.
By focusing on occupation-specific changes, we can even compare similar workers doing similar work with different exposure to occupational decline. For example, we can distinguish typists, whose employment fell sharply, from secretaries, whose employment grew. We assemble data on forecasts on occupational employment growth from the 1980s, which allow us to compare workers in occupations that were expected to grow similarly, but differed in their actual growth. This allows us to evaluate the effect of occupational decline on those who were exposed to it, relative to those who were not.
We find that occupational decline reduced workers’ career earnings (over 28 years) by around 2-5 per cent. About half of this earnings loss was due to reduced time spent in employment, and the rest was due to lower earnings during periods of employment. We also find that employment losses for those facing occupational decline were partly accounted for by increased time spent in unemployment and, to a lesser extent, government retraining. For older workers, occupational decline also brought about slightly earlier retirement.
The losses from occupational decline were not, however, evenly spread. Workers whose initial earnings were low compared to others in the same occupation suffered larger losses of around 8-11 per cent of their career earnings. These workers also spent more time than others in both unemployment and retraining.
We also find that earnings losses suffered by workers in declining occupations were larger during recessions. This highlights the importance of labour market resilience in cushioning the blow for workers facing occupational decline.
Our findings suggest that occupational mobility is important in mitigating the losses from occupational decline. Workers who began their careers in occupations that later declined were less likely to remain in their starting occupations than other workers. This higher mobility away from declining occupations reduced labour supply for declining jobs and likely helped shield those who remained in declining occupations from heavier earnings losses.
Occupational decline may arise from various sources, including changes in preferences and international trade, but in the long run technology is an important driver of occupational decline. To study its role we repeat our analysis focusing only on occupations with declines that are more directly attributable to technological changes, such as computerisation, the rise of modern communication equipment, and the spread of robots. We find that these technology-driven occupational declines have a similar impact on workers’ career earnings as other occupational declines.
We also explore whether our findings are specific to Sweden, with its particular set of labour market institutions. We repeat our analysis using data on a smaller number of US workers. Again we find that average earnings and employment losses from occupational decline were small, although our estimates for the US are admittedly less precise.
Our findings suggest that earnings losses from occupational decline are smaller than the losses suffered by workers who experience mass layoffs, as reported in the existing literature. This is likely because occupational decline is typically gradual, and can be partly managed through retirements, reduced entry into declining occupations, and increased job-to-job exits to other occupations. Gradual occupational decline may also have a less pronounced impact on local economies compared to large, sudden shocks, such as plant closures.
Nevertheless, there are reasons to believe that future occupational decline could still have a significant negative impact for workers. First, new technologies such as machine learning may develop more quickly than those in the past, causing more rapid occupational replacement, making adjustments more costly. Second, future occupational decline may impact occupations where workers have invested more heavily in occupation-specific skills or benefit from economic rents. In this case the earnings losses from future occupational decline may be higher than those we estimate.
Finally, and perhaps most importantly, our findings show that low-earning individuals are already suffering considerable earnings losses, even in Sweden, where institutions are geared towards mitigating those losses and facilitating occupational transitions. Helping these workers stay productive when they face occupational decline remains an important challenge for governments.
Stay updated with all the insights.
Navigate news, 1 email day.
Subscribe to Qrius